Dictionary-based Pathology Mining with Hard-instance-assisted Classifier Debiasing for Genetic Biomarker Prediction from WSIs

2026-03-26

AI Summary

Accurate prediction of genetic biomarkers—such as microsatellite instability (MSI) in colorectal cancer—from whole-slide images (WSIs) is hindered by the difficulty of modeling complex pathological interactions and the risk of overfitting due to abundant irrelevant regions. To address these challenges, this work proposes D2Bio, a novel framework featuring a dictionary-based hierarchical pathology mining module that transcends local neighborhood constraints to capture global, fine-grained pathological interactions. Additionally, D2Bio introduces a bias-mitigation mechanism via a hard-example auxiliary classifier that requires no additional annotations, focusing learning on task-relevant difficult instances to yield unbiased representations. Evaluated across five independent cohorts, D2Bio achieves an AUROC improvement of over 4% compared to the best existing method on the TCGA-CRC-MSI cohort, while demonstrating strong generalizability, clinical interpretability, and potential for survival analysis.

Abstract

Prediction of genetic biomarkers, e.g., microsatellite instability in colorectal cancer is crucial for clinical decision making. But, two primary challenges hamper accurate prediction: (1) It is difficult to construct a pathology-aware representation involving the complex interconnections among pathological components. (2) WSIs contain a large proportion of areas unrelated to genetic biomarkers, which make the model easily overfit simple but irrelative instances. We hereby propose a Dictionary-based hierarchical pathology mining with hard-instance-assisted classifier Debiasing framework to address these challenges, dubbed as D2Bio. Our first module, dictionary-based hierarchical pathology mining, is able to mine diverse and very fine-grained pathological contextual interaction without the limit to the distances between patches. The second module, hard-instance-assisted classfier debiasing, learns a debiased classifier via focusing on hard but task-related features, without any additional annotations. Experimental results on five cohorts show the superiority of our method, with over 4% improvement in AUROC compared with the second best on the TCGA-CRC-MSI cohort. Our analysis further shows the clinical interpretability of D2Bio in genetic biomarker diagnosis and potential clinical utility in survival analysis. Code will be available at https://github.com/DeepMed-Lab-ECNU/D2Bio.

Problem

• genetic biomarker prediction
• whole slide images
• pathology representation
• model overfitting
• microsatellite instability

Innovation

• dictionary-based pathology mining
• hard-instance-assisted debiasing
• genetic biomarker prediction
• whole-slide image analysis
• classifier debiasing